Mizuno reports increased excess heat



    Jed: you have posted the same equations I posted - from which I derived Umax = 1.23V (at Re = 12,000). I am happy to go through these equations with you if you doubt this. That gives an average velocity 19% less than the middle of pipe peak velocity.


    So I cannot agree with your bound - if it is based on the equations.


    If based on the measurements: I'm trying to work out how you use the CW-60. Have you personally used it - you would know. The spec I've found is


    http://custom.company.weiku.co…r-Heat-Wire-14658647.html


    The hot wire is as you know in a probe. Could you tell me how you insert the (long) probe in the pipe? Is it meant to measure flow perpendicular to the probe, or flow parallel to the probe, or does it make no difference?


    From the anenometer data I get an accuracy of approx +/- 0.3m/s, which corresponds to +/- 8.5% on speed, and therefore power.


    So my bound currently would be +0 - 19% (airflow velocity profile) +8.5% - 8.5% (anenometer accuracy).


    That gives +8.5% - 26% bounds overall on airflow from the measured result.


    The -19% from the air velocity profile can maybe be tightened. I'm just not sure because of the big difference between the equations for turbulent flow and the very flat measurements - which seems anomalous. While there is probably some good explanation I'm not confident.



    RB: forgive me but do you just skip over my posts? I posted the velocity profile 3 posts above: Mizuno reports increased excess heat


    Also note that is turbulent flow power law velocity profile with n = 5, typical of Re = 10,000. (The profile varies slowly with Re, so 11,633 is very close to this).


    You can get another approximation using the equations I posted earlier - again standard turbulent flow stuff. It is quite similar.


    You are no doubt wondering about the probe measurements of velocity, which are so very flat. Maybe that is because the flow is too near the blower to be typical turbulent flow in the pipe. Or maybe the probe is disturbing the flow velocity profile. I am not sure, and when not sure I prefer to have larger error bounds rather than assume there cannot be any problem.

    Quote from RobertBryant

    what velocity profile does THHnew use to get likely 24 % error?

    Quote from THHuxleynew

    I'm just not sure because of the big difference between the equations for turbulent flow and the very flat measurements


    What is the big difference


    What are the very flat measurements you are using,


    Please show numbers.

    Show what you predict the measurements should be versus what you predict using equations


    State all assumptions and equations




    Not words

    Anyway - I am trying to work out the best possible bounds I can. So far Jed is proposing +0 - 17%. But I can't agree with this, he is not including the anenometer accuracy. Also, I think 17% is a guess. If the profile is truly as flat as measured we get get less than this. If something about the probe flattens the profile, and it is typical of turbulent flow in a pipe at this speed, we have ~19%. For a bound, rather than a guessed typical value, I'd want to add 5% to give 24%. So for a total bound I'd be happy with:


    19% (from turbulent flow velocity profile) 8.5% (from anenometer accuracy at 3.5 m/s) 5% because the turbulent flow stuff is uncertain, for safety.


    I then get a safe bound of -30%. The positive bound is much simpler, +8.5%.


    Maybe we can tighten the +/- 8.5% bound. here is anotehr anenometer datasheet different from the one linked above:

    https://axel-search-e.as-1.co.…8-01/?print=true&pdf=true


    The accuracy spec is inconsistent: here given as 5% rdg +/- 1 digit. At 3.5m/s +-1 is roughly 3% so we get +/- 8% from this - about the same even though the spec is different (no full scale referenced accuracy spec).


    Jed: what accuracy did you reckon for the anenometer - and did you get this from datasheet?


    RB - sigh. You refuse to read my previous posts so I have to repost things many times.


    Flat measurements: Fig 9 in first M paper (eyeballed at no more than 10% difference across tube between centre and 3mm from edge)

    The standard flow: see my picture above, or note u = Umax(1 - r/R)^(1/n) where n = 5 from Re=10,000. 3mm from edge would be roughly 50% difference (eyeballing graph) or 39% difference (from equation). e.g 61% of max value.

    Or take the slightly different equation I linked earlier (last page I think).

    Or look up standard turbulent flow in pipe stuff yourself?

    Anyway: since some people may not remember - I want to point out again:


    • This issue is irrelevant if you trust the sample R20 results. That trumps any R19 results.
    • It is also irrelevant if you trust the control vs active results for R19, sure that the conditions under which control and active measurements are made do not differ.
    • It is however relevant if you rely on the absolute calculation, or if you rely on the calorimeter efficiency (heat loss) being as stated in the paper.


    RB - I make no claim to academic credentials here, so there is no point using them. Nor would it make my (quickly written for fun) stuff here more reliable if I had them.


    I get 11,633. However you can see from the equations that the difference between 10,000 and 11,633 is very small. On a previous page I gave the value from a different approximation, using 11,633. And previously I've used 12,000 as a ballpark Re. It makes no real difference which you choose, they are all fully developed turbulent flow.


    However you read that, so you know.


    In the same post I derived the 11,633 from two web calculators, which I linked (I thought some here would appreciate that). The initial data is 3.5m/s, (or maybe 3m/s, I can't remember what I chose), 40C, air, pipe diameter = 66mm.


    Do you remember that post, written earlier today, or do I need to direct you to it again?


    BTW - from your contribution here I doubt you are slow at Fluid Mech. However you seem unwilling to investigate questions of interest - and hostile when I do so.

    Dear mods. I do apologise. I have laid out the very interesting issues around how Mizuno airflow can be bounded, what are the key error sources, what is the issue about velocity profile. I've spent some time doing this in detail today. I'm sure others like RB here can take my posts and check them. I've given references. Jed can also comment on the anenometer accuracy issue.


    But I find I am becoming impolite - something that happens when others make personal replies to my posts instead of looking at the data for themselves.


    That is not what I want to do. However I truly think I'd need to be a saint not to so so - and I am no saint.

    Quote from THHuxleynew

    That is not what I want to do. However I truly think I'd need to be a saint not to so so - and I am no saint.

    but I still can't see how you get likely 24% error without assuming retrograde velocity in the annular region


    I did notice that the velocity spreads in the Mizuno data get less as the velocity increases


    at v=5.0 m/s I get via Etoolbox an even higher ~ Re 21,000


    this is consistent with a flatter profile at higher velocities


    A full examination of this matter will require more than eyeballing


    In the end it is rather futile ...although heroic considering it is unpaid effort


    As I have pointed out an error of 10% or so systematic in both calibration

    and active runs makes little difference to the COP.


    The object of the calorimetry was to find the best COP


    I guess it gave some direction towards finding the R20 's huge COP


    That looks OK to me. Your number is a bit higher than mine because:


    (1) you chose a higher velocity (4 m/s). I chose the middle point of that graph: 3 m/s.

    (2) you are using 60F temperature (16C) not 40C figures for density and kinematic viscosity: temperature changes the result a little.


    From my post here you can get the figures needed for 40C, and compute the result for whatever velocity you like.


    From same post you can see the effect:


    Umax = peak middle of pipe velocity

    V = average airflow velocity


    Umax = V(1 + 1.33*(100Re)^(-0.125)) (that is one of the two commonly used approximations).


    so changing Re from 10,000 to 20,000 changes the multiplier by 2^(-0.125) = 0.9. So Umax / V goes from 1.24 to 1.22. It is not at all sensitive to the exact Re (as long as it is large enough to stay fully turbulent).


    -----------------------------


    There is another approximation for velocity profile in turbulent flow - with very similar results


    This approximation comes from the power law velocity profile working out n from Re - but again n does not change much with Re 10,000 or 20,000 and the result is similar:



    v = Umax(1-(r/R))^(1/n)



    V = (integrate this * 2*pi*r) and divide by pi*R*R)


    Sorry - I actually did this integration a while back but have lost the will to do so now: and can't immediately find a reference that provides av velocity from power law doing this, although I know there is one somewhere...


    Anyway the alternative approximation above gives similar (though not identical) results.

    but I still can't see how you get likely 24% error without assuming retrograde velocity in the annular region


    I did notice that the velocity spreads in the Mizuno data get less as the velocity increases


    at v=5.0 m/s I get via Etoolbox an even higher ~ Re 21,000


    this is consistent with a flatter profile at higher velocities


    If the anenometer data represents the real velocity profile, then:

    (1) it limits the difference between peak and average velocity to around 10% (not 20%) - I'd guess. Difficult to be sure from how the measurements are presented.

    (2) It is wildly inconsistent with any of the typical turbulent velocity profiles: and note there is no significant different between Re=10,000 and Re=20,000.



    I can think of two solutions:


    (1) The measurements are being done close enough to the fan that the fan turbulence determine the velocity profile and this has not had tome to settle into the "typical" pipe profile.

    (2) The probe, when inserted into the pipe to perform these measurements, changes the profile by adding turbulence.


    I cannot tell which of these is true, or if there is some other solution I have not thought of.


    Also it is worth noting that the data on velocity profile in the paper is very difficult to read, because shown graphically as different shaped dots. Jed however maybe has the numerical data from which this comes.


    Given this uncertainty taking the theoretical value in pipes as a bound is sensible, even though a more accurate bound might be found after more work.

    @THHuxlynew, your calculations seem to be erroneous. Here is a copy of my fluid dynamics textbook:


    For Equation 36, you use f-(100Re)^(-⅛) and that gives the ratio of Umax to V of 1.24 but if you use the correct equation that reduces to only 4%. Also note one more important point. The pipe diameter is 6cm and length estimated to be about 30cm so about 5xD only so the flow is nowhere near being fully developed. Even the 4% figure which gives for something like 60xD or at least starts to get close there will be much, much less for a distance only 5xD from the entrance.


    In conclusion, the flow error due to Umax:V ratio is <<4% and probably <2%. To this you only need to add the uncertainty of velocity measurement from the actual anemometer used.


    QED


    Daniel, thanks for this.


    You make two separate points.


    The first is wrong. Equation 36 gives that as the value of f= (100Re)^(-0.25) but Equation 39 uses sqrt(f) which therefore changes (100Re)^(-0.25) into (100Re)^-0.125).

    QED. 20% rules.


    The second is interesting, and I've proposed it myself above as a possible solution. Indeed the flow may not have fully equilibrated to the 66mm pipe boundaries, in which case it will be dependent on the blower etc.


    https://en.wikipedia.org/wiki/Entrance_length


    10*6.6 = 66cm so I (now, having found the info on entrance length) agree with you. The pipe is not long enough.


    Based on that I'd reduce my bound of the average / peak velocity to: 0 to -10%.


    Giving a bound on the airflow of +8.5 to -18.5% that measured. (The 8.5% comes from the anenometer accuracy).


    The paper says that the tube is there to make the airflow velocity profile more consistent. A bit of amplification there (e.g. that the turbulent flow from the blower does not have time to develop in the tube) would be helpful.


    Best wishes, THH

    Yes I see my math was wrong and you are correct in the first point. Please accept my apology.


    As for the flow development I agree the difference from Umax to V is going to be less than 10%. If I had time to run this on FLUENT, I could give you an exact number but if you study the charts available, the development of the flow profile in 5xD is going to be very low and that's why I agree with the <10% figure. Perhaps if I have time next week I can run the simulation but the air coming out of the fan is going to be very turbulent and the measurement point at 5xD is not going to allow the flow to develop at all. I think based on this we can agree that the COP figure given in this experiment is still far beyond any possible error in the flow measurement. Right?


    Output heat was stabilized at 303W @ input of 50W so if we assume the worst case scenario, and multiply the 303W measured output by 0.815 we get 247W so the lower bound for the COP reading would be 247/50=4.94, still far and above anything that could be explained chemically, especially if this reaction continued for days as Mizuno is showing.


    Just for fun, if we assume that the flow is both laminar and fully developed, Avg. flow is going to be ½ of Umax and then the measured 303W (ignoring calibration) would become 150W, which still gives a COP of 3 which is still very good by LENR standards so I guess I don't really understand why we are spending so much time on this issue. The measured COP is so high, the result is still robust in light of all the possible errors is it not? Or am I missing some part of your argument?


    Oh and the empirically measured data all the way up to 3mm from the outside wall seems to follow the undeveloped flow theory so can we agree that this was a good exercise, but any such flow error would only reduce the COP from the reported 6 all the way down to 3, all other things being equal and although this is nowhere near being valid from the empirical data, even giving you the maximum benefit of the doubt, the COP is still far above anything that could be accounted for through such errors.

    I think based on this we can agree that the COP figure given in this experiment is still far beyond any possible error in the flow measurement. Right?


    Yes, agreed, beyond. We have gone from +50% to +(1,5*0.915*0.9) = 1.23 +23%


    That is still beyond any obvious error in the input power measurement, or heat measurement.


    However, things are not quite so simple. The data given in the paper is corrected for estimated calorimeter efficiency (I believe). If I'm wrong and these are non-adjusted results, then I agree entirely with you.


    Thus if I've understood it correctly Figure 3 in the later paper shows a graph for "heat recovery" which plots calorimeter efficiency against reactor temperature, and shows for the high temperatures corresponding to COP = 1.5 a recovery of 78%.


    That presumably means that the actual output power bound as measured is 1.23*0.78 = 0.96


    The errors we need now to consider are therefore:

    • Calorimeter calibration error (not sure how to bound this other than with calorimeter efficiency?)
    • Temperature measurement error. Not known without RTD spec, the paper indicates they are calibrated against each other, say 2%, unless there are issues to do with the enclosure heating up and changing conducted heat to the RTD from its mount?
    • Error due to room temperature change over experiment (needs to be properly bounded, or measured during experiment)


    The calibration error is dominated by heat loss through the calorimeter walls, and paper says that this is re-measured before each run. I'm not sure if we have the data for this, or how it is processed. More importantly, in this setup, calibrations and active runs use a different reactor, placed in a different position in the enclosure, so in the two cases the heat losses could be very different. Any differences introduce an error and we do not know what this is.


    My gut tells me that none of these errors are likely to be so large as to invalidate the excess heat results.

    My brain tells me not to trust my gut, since figures when worked out properly often surprise,

    Quote

    I have never read you being so wrong, on so many counts Jed. All I can say is that you are letting your ego get in the way. I am just an interloper, put in a position to make a difference. I am going to do that to the best of my ability, and I know many others will help. I was hoping you would be a part of that. Guess I was wrong.

    Are you saying you are planning to kidnap Mizuno and take him to ICCF in your luggage?

    Quote from JedRothwell

    Of course you know what it is. Of course you can give bounds. It is right there in Fig. 4. The velocity is uniform across 60 mm of the 66 mm orifice. The velocity cannot possibly drop abruptly to zero in the last 3 mm.


    There were velocity profile equations in textbooks.. accurate to within 10-15%??

    I remember using them.... profile depended on smoothness..

    friction.


    thirty years ago they were there... perhaps they have been lost in this era of software


    perhaps I can find them by August.:)

Subscribe to our newsletter

It's sent once a month, you can unsubscribe at anytime!

View archive of previous newsletters

* indicates required

Your email address will be used to send you email newsletters only. See our Privacy Policy for more information.

Our Partners

Supporting researchers for over 20 years
Want to Advertise or Sponsor LENR Forum?
CLICK HERE to contact us.